Research Update

Soil sensor data

Developed generalized script to read in soil sensor data from logger output.

source('code/read_soil_data.R')

Summarized these data in various ways. Here, by site, aspect, day and variable.

soil_data <- soil_df %>% 
  group_by(site, aspect, date, variable) %>% # Change site for aspect
  summarize(dayMin = min(value, na.rm = T),
            dayMean = mean(value, na.rm = T),
            dayMax = max(value, na.rm = T)) 

Time series of the daily-summarized data. Bands of color across the plot indicate windows of missing data, when sensors were damaged by wildlife. We are mostly missing data from the flat sites, which is fortunate.

Distributions of measurements by site and aspect from summer months (June-September), 2017 and 2018 together. Differentiation by aspect is not consistent across sites, but some trends emerge. Buffalo is a much warmer, drier site. Maple has highest soil moisture.

Seedling survival and growth

source('code/seed_frames_analysis.R')

Probability distribution of PSME measurements. Not much differentiation among aspects when lumping the whole growing season.

Probability distribution of PICO measurements. Not much differentiation among aspects when lumping the whole growing season.

Counts and proportions of PICO over the growing season. In general, seedlings died during the summer. Pattern by aspect not as clear.

Counts and proportions of PSME over the growing season. In general, seedlings died during the summer. More gradual pattern of decline compared to PICO. Pattern by aspect still not consisten among fires.

Seedling allometry

source('code/seedling_allometry.R')

Built a model of seedling allometry, based on the equation we used in EcoPhys. It models total biomass as a funciton of log(basal diameter ^2) and log(height). See below for coeffecients, R2, etc.

summary(ecophys_model)

## 
## Call:
## lm(formula = log(total_mass) ~ log(basal^2) + log(height), data = allometry_df)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -0.88483 -0.15093  0.06642  0.20764  0.91014 
## 
## Coefficients:
##              Estimate Std. Error t value Pr(>|t|)    
## (Intercept)   -8.4601     1.1826  -7.154 4.04e-08 ***
## log(basal^2)   0.7327     0.1687   4.344 0.000132 ***
## log(height)    1.5426     0.3313   4.657 5.38e-05 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.4246 on 32 degrees of freedom
## Multiple R-squared:  0.8662, Adjusted R-squared:  0.8578 
## F-statistic: 103.6 on 2 and 32 DF,  p-value: 1.057e-14

Model residuals look pretty good, although the sampling is uneven (because I favored small seedlings that were closer in size to mine).

Next steps

• Look at individual seedlings, for inference, and to clean out any obviously problematic measurements (e.g., seed caps). However, it will be very difficult to differentiate legitimate measurement errors from accidental ones.
• Use allometric model to assess change in total biomass over time (instead of height or basal).
• Look at change in biomass over time as a function of temperature, moisture, elevation (instead of the proxy I am using now, aspect). This step will ultimately lead to building a model that predicts survival and/or growth based on a combination of the abiotic variables we measured.
• Obtain a product(s) that describes the key abiotic variables across the study area, OR somehow extrapolate our measurements of based on relationship with aspect, elevation, topo position, etc.
• Estimate a spatial model of survival/growth based on these products (perhaps incorporating our measurements).